Objectives: To compare different forms and ratios of serum prostate-specific antigen (PSA) to determine which form or ratio provides optimal diagnostic specificity and sensitivity in distinguishing between benign prostatic hyperplasia (BPH) and clinically localized prostate cancer.
Methods: Serum samples were obtained from 47 patients with BPH and 39 with clinically localized prostate cancer. Patients with BPH underwent either transurethral resection of the prostate or transurethral microwave thermotherapy. Patients with prostate cancer, all of whom had no metastases on radionucleotide bone scans and no pelvic lymph node involvement, underwent either radical external beam radiation therapy or radical retropubic prostatectomy. All patients had pretreatment serum PSA levels between 1 and 20 ng/mL. The different forms of serum PSA (free PSA [PSA-F], PSA complexed to alpha 1-antichymotrypsin [PSA-ACT], and total PSA [PSA-T]) were measured using different monoclonal antibodies against PSA and ACT and immunofluorometric assay techniques. Furthermore, three ratios (PSA-F/PSA-T, PSA-ACT/PSA-T, and PSA-F/PSA-ACT) were calculated.
Results: By receiver operating characteristic curve analysis, the performance of the different forms and ratios were compared. The PSA-F/PSA-T ratio had the greatest area under the curve (AUC, 0.776), significantly larger than that for PSA-T (0.612; P = 0.024). For PSA-ACT/PSA-T, the AUC was 0.695 (P = 0.283 versus PSA-T) and 0.773 for PSA-F/PSA-ACT (P = 0.051 versus PSA-T). At a cutoff level < 0.17, PSA-F/PSA-T had a sensitivity of 79%, a specificity of 66%, and a positive predictive value of 66% compared with 74%, 38%, and 50%, respectively, for PSA-T at a cutoff level > 4.0 ng/mL.
Conclusions: The PSA-F/PSA-T ratio gives the best diagnostic performance compared with that for other forms and ratios of PSA and will reduce the number of prostatic biopsies in patients with BPH.